Systems and methods for tuning light sources for use with object detection algorithms

An object detection system, comprising:

• a lighting system having one or more luminaires to illuminate an object, the lighting system configured to selectably output one or more light output settings, and the light output settings includes at least relative intensities of wavelengths of light;
• at least one image sensor positioned to obtain an image of the object;
• and
• at least one processor coupled to the image sensor to receive the image of the object, the at least one processor comprising:
• a monitoring engine configured to process the image using deep learning-based object detection and configured to determine if the image has an expected confidence value resulting from the processing of the image from the deep learning-based object detection that meets a predetermined threshold, the confidence value including a measure of contrast, intensity and/or uniformity of frame containing an image of the object, the confidence value being indicative of a likelihood of detection of the object depicted in the image by the deep learning-based object detection;
• and
• a light settings calculation module configured to select the light output settings of the lighting system to improve the expected confidence value to meet the predetermined threshold using a plurality of rules if the monitoring engine determines the image does not have the expected confidence value to meet the predetermined threshold, the light settings calculation module being configured to add to or amend the plurality of rules over a period of time, and the light settings calculation module having a multichannel color mixing algorithm to compute duty cycles for each color channel of each luminaire to select the relative intensities of wavelengths of light;
• and
• a training module configured to learn confidence values corresponding to selected object properties, the monitoring engine being further configured to obtain a confidence value from the training module for the image, the training module having a first data structure including confidence values for an array of properties of a plurality of objects, the confidence values in the training module including a plurality of values corresponding to different lighting conditions for each of the selected object proerties,
• wherein the light settings calculation module is configured to select the light output settings of the lighting system to improve the expected confidence value to meet the predetermined threshold using the plurality of rules only if the monitoring engine determines the image does not have the expected confidence value to meet the predetermined threshold.